The Research Of Weak Signal Extraction Method Based On Generalized S Transform In Seismic Data

As seismic exploration gradually develops into deep complex tectonic lithologic reservoirs,the obtained seismic data has low signal-to-noise ratio and low resolution,the effective signal is submerged in the background of strong noise.The denoising technique in conventional data processing is not suitable for data processing with low SNR in deep strong noise background.The low SNR seismic data processing requirements reasonably effective denoising but do not affect the extraction of weaker amplitude signals.Therefore,it is necessary to develop new denoising techniques and methods based on conventional seismic data denoising methods.It is suitable for the identification and extraction of seismic weak signals in the background of strong noise.In this paper,the nonlinear and non-stationary features of seismic signals are suitable for processing using time-frequency analysis methods.Through comparative analysis,the applicability of traditional time-frequency analysis method,S-transformation and generalized S-transformation(GST)to non-stationary signal processing is studied.It shows that S transform and its GST have better time-frequency aggregation and are more suitable for seismic signal processing.Aiming at the shortcomings of the fixed change trend of the S-transform time-frequency analysis window function,an optimized S-transform with a good time-frequency focusing hybrid control Gaussian window function is introduced.Based on this optimized GST a series of strong noise background weak signal recognition and processing methods are developed.The principle and calculation process of the frequency-domain adaptive filtering denoising algorithm based on GST,adaptive seismic data processing method based on GST and Gaussian smoothing,Combined GST and CEEMD seismic weak signal processing method and surface wave suppression method based on joint GST and TT transform is also given.And through time-varying signals,synthetic seismic records and actual seismic data test analysis and it proves that this series of methods has good applicability to low SNR seismic data processing.It has the ability to identify and extract effective weak signals in a strong background with strong noise.Aiming at the shortcomings of different algorithms in this series of methods in dealing with low SNR seismic data,different optimization methods to be improved are also given.